Sign in with
Sign up | Sign in
Your question

Power Supply Efficiency Myths

Last response: in Components
Share
a c 271 ) Power supply
August 6, 2012 2:48:10 PM

There is a lot of confusion and many myths related to power supply(PSU) efficiency, first and foremost is not knowing what exactly efficiency means. In the context of computer power supplies efficiency refers to how well the power supply is able to convert the AC power it recieves from the wall into DC power it feeds to your components. It is important to know how this actually applies to the amount of power that enters and leaves the power supply; a power supply rated at 500 W can deliver 500 W of power to the system regardless of its efficiency, the efficiency tells you how much power it must draw from the wall to deliver those 500 W it does not affect its output capacity, a 500 W power supply that is 90% efficient can deliever the same amount of power as one that is only 80% efficient. If the power supply is 80% efficient it needs 625 W(500/0.8) from the wall, those extra 125 W are turned into heat by the power supply, while a 90% efficient unit would only be drawing 555 W from the wall, meaning it is dissipating only 55 W as heat, or only 44% of the heat the 80% efficient unit was creating.

Myth #1 is that using an oversized power supply will consume significantly more power at idle than a smaller unit since PSUs are less efficient at low loads and high loads, and are at their best between 40% load and 80% load. This is true, a power supply is less efficient at low load and there is a "curve" if you plot a PSU's efficiency over a variety of loads and zoom in real close by truncating your Y axis on the graph; however, if you look at the graph from 0 to 100% efficiency it is a pretty flat line between 10% and 100% load for most units with 80 plus basic certification or better, and the graph doesn't plummet as it approaches 0% load as many people draw it, even tom's has made this mistake before.


If we look at actual plots of 11 different power supply's efficiency at 10%, 20%, 50%, and 100% load we can see just how much changing the scale can skew our perception. There are 2 80+ platinum, 3 80+ gold, 3 80+ bronze, 2 80+ basic, and a single unit(Corsair 430CX) with no 80+ certification. You will notice that almost all of the 80+ rated units are very flat and none of them have a low load efficiency below 74% which is still pretty good for units that can output up to 1375 W.

Here is the plot nicely zoomed in to provide a nice look at a curve which gives the illusion that there is a significant change between maximum and minimum efficiency levels

Now here is an honest graph to show what it really looks like, you will notice that there is no massive drop between maximum and minimum efficiency levels, it is all pretty flat and not really a curve.


Even the Corsair 430CX which has no 80+ certification does not show a significant curve and it has efficiency that varies from 70% to 83%. Almost all power supplies you can buy today have a fairly flat efficiency curve, especially if they have any version of 80+ certification.

There are still some units out there that have abysmal efficiency across the board. For example, the Cooler Master eXtreme Power Plus series which still seems to be common unfortunately. Hardware Secrets did a review of a couple of models from the series to show their efficiency across the board, and they too had a fairly flat line; however, it was pretty miserable efficiency and they never did break 80%.
eXtreme Power Plus 600 W eXtreme Power Plus 550 W

Moral of this story? Make sure you get a unit with at least 80+ basic certification, the ones without it can be anywhere from absymal to decent, and never great.


Myth #2 is that the PFC(power factor correction) present in the power supply has an effect on its efficiency one way or another.There are two kinds of PFC, Active and Passive. Active PFC uses electronics to accomplish its job more effectively than passive PFC which just uses a big coil, but being active it consumes some power which does hurt efficiency a tiny amount. However, active PFCs tend to boost the voltage inside the PSU up to a higher level than units with passive PFC do, which allows the rest of the power supply to run slightly more efficiently, so the effects pretty much cancel each other out. In this day and age, passive PFC units(ones with red voltage selector switches) should not be purchase, they are old designs and modern active PFC units can handle a much larger range of incoming voltages. Having Active PFC also reduces the load that the power company sees but that is a story for another day.


Myth #3 is that 80+ was meant for individual consumers; while an 80+ rated unit is more efficient and will save you some money the long run it wasn't meant to save much money on the consumer level, it has much larger effects at the corporate and industrial level. If you are a company with small office machines that use 200 W under load and they are under load 8 hours a day, 5 days a week, 50 weeks a year they will consume a significant amount of power. At 70% efficiency(the level most old PSUs are at) they will consume 285.7 W from the wall, at 82% efficiency(bronze) they will consume 243.9 W from the wall, and at 87%(gold) they will consume 222.2 W. For a single machine this doesn't seem like much, but for 1000 machines in an office that means we are talking about 444.4 MWh for the 80+ gold office, 487.8 MWh for the 80+ bronze office, and 571.4 MWh for the office using old power supplies, that is over a 100 MWh difference each year which obviously saves money for them on electric bills, but also reduces the load on their air conditioning system, and reduces the generating capacity needed at the power station so you have actually just saved that energy three times over. We also need to remember that power supplies aren't spectacularly efficient, a high efficiency gas power plant is round 60%, in order to provide 571.4 MWh would need to burn gas with an energy of 952.3 MWh, while it would only need 740.6 MWh to power the gold rated office, that is a 222 MWh difference, those 222 MWh are getting dumped into the atmosphere as raw unused heat along with all the exhaust fumes of the power plant, that is a massive quantity of waste energy that can be saved with higher efficiency power supplies.

We as human's tend to think on a small local scale, but for engineering purposes, it is the massive global commercial scale where you can have the most impact and that is what 80+ was meant for.


Sources:
Efficiency data for the charts were gathered from jonnyguru.com review ids 214, 231, 239, and 292-299
a b ) Power supply
August 6, 2012 4:12:17 PM

great info bro!
thanks.
a c 248 ) Power supply
August 6, 2012 4:41:01 PM

Great post!

I knew about myth #3 because I read the 80+ mission statement the very first time I visited their site a few years ago. It was clear that 80+ was meant for enterprise.
Related resources
a b ) Power supply
August 6, 2012 4:49:12 PM

i have a little doubt,i hope someone can help me-
HX series from Corsair are rated as *Silver* certified but i heard they achieve Gold standards? how is that possible? if it's true then why corsair is selling them as Silver certified units?
a c 271 ) Power supply
August 6, 2012 5:03:24 PM

While confusing it is true, the hx 750 and 850 do meet gold level efficiency and are certified by 80the plus as goldto but just barely and cannot quite hold those levels at higher temperatures so rather than advertise them as gold and get dinged in reviews that test units in a hotbox they marketed them as silver and passed reviews with flying colors.
a b ) Power supply
August 6, 2012 5:10:30 PM

got it,thanks!
a b ) Power supply
August 6, 2012 5:26:43 PM

Great explaination.
On active verse passive, Both can reduce effiency, But the acitive - more so than pasive. The Input Impeadeance to a PSU is primarily capacitive, the Coil simple counters this. The Loss is do to the DC resistance of the coil. As the old saying goes you can not get sometihng for Nothing, so a passive system does consume a small amount of power. Both function to reduce the Power factor and thus reduce the Peak current. Reason power companies love a PF of 1 is that it reduces the IR losses over the Transmission lines, thus reducing the Power generation requirements.

Ref PSU size, You hit the nail on the head for the average home user eff difference for most PSU may equate to a cup of java a month. On large PSUs, I still recommend You s limit the max size so that Idle power is at least 10 %., Provided you can still stay below the 80% at max load. Reason for the 80% is more a longevity issue than cost savings. By design Very low power is hard on PWM circuits employed in Switching PSU. In fact the Initial Computer Power supplies would go up in smoke if Power output was zero (NO LOAD) - after a few years this was corrected (initially added resistor to provide a bias current - Then added a safty circuit to simply switch PSU off when No load is detected.. May be wrong on this but when the duty cycle becomes close to zero current accually goes very high
a c 271 ) Power supply
August 6, 2012 11:18:25 PM

I was trying to avoid getting too technical as that also increase the likelihood someone can find a point to nitpick. For active versus passive, for the most part the ESR of the coil is very low so it is a minimal power loss and negligible for purposes other than design of the system, in the same way that the Active PFC circuit has minimal power losses, for the most part they aren't going to be enough to affect the efficiency when you have a few hundred watts at play and that is what i was going for, i tried not to get stuck on the little things in this scenario but anyone with a significant electrical background should be able to understand that passive will still consume some power, just not much.

I try not to scare people with complex and reactive power, no matter how much i love using phasors because they make AC analysis super easy, they still confuse the average person we have here, but you are right, the closer PF is to 1 the less the power company has to generate to make up for line losses which is why 80+ requires at least a PF of 0.9 for bronze and higher. I believe that there are many places outside of north america that actually have to pay for apparent power not real power so the PF of the equipment is a relevant concern for residential users in other parts of the world, but i have yet to figure out a good way to explain PF that doesn't involve using my circuits II textbook.
a b ) Power supply
August 7, 2012 1:25:56 AM

Great read.

Myth #1 explained = is what I always thought.
a b ) Power supply
August 7, 2012 9:41:10 AM

another question(s)-
1-what is the difference between multi rail and single rail power supply?
2-which one is better? a) for single gpu setup. b)multi gpu setup.

thanks!
a b ) Power supply
August 7, 2012 1:22:34 PM

^it took me almost 20 minutes to understand it lol but i guess it was worth it.
another one please-
is modularity in a power supply very beneficial? from cable management to airflow.
i mean is there a strong reason to spend 20 bucks extra for a modular power supply?
thanks.
!